An Intelligent LLM-Based Mental Coping Detection Framework.

Area of Science:

• Computational Social Science
• Mental Health Informatics
• Natural Language Processing

Background:

• Social media analysis for mental health (MH) often prioritizes disorder detection over coping mechanisms.
• User-generated social media text offers MH signals but lacks clinical validation.
• Existing approaches overlook the nuances of psychological coping strategies.

Purpose of the Study:

• To develop and evaluate a two-stage framework for detecting mental coping strategies in social media text.
• To integrate user-identified coping strategies with expert-validated knowledge bases.
• To generate clinically relevant and evidence-based responses for mental health support.

Main Methods:

• A two-stage framework combining a coping strategy classifier and a Retrieval-Augmented Generation (RAG) system.
• The classifier identifies coping strategies from user-generated text.
• The RAG system generates responses grounded in expert knowledge and classifier output.

Main Results:

• The coping classifier achieved 79.7% accuracy and a 0.797 weighted F1 score on a held-out dataset.
• The RAG system generated structured, readable, and expert-verified responses.
• The classifier-conditioned RAG architecture successfully grounded responses in validated psychological knowledge.

Conclusions:

• The proposed framework offers a scalable method for analyzing mental coping strategies on social media.
• It bridges the gap between large-scale social media data and evidence-based psychological insights.
• This approach enhances mental health informatics by incorporating validated coping mechanisms.